Follow Up Letter to California Attorney General on OpenAI’s Nonprofit Status
By Robert Weissman
RE: OpenAI, Inc.
Dear Attorney General Bonta and Deputy Attorney General Lamerdin:
This letter is to follow up on my letter of January 9 encouraging you to investigate OpenAI, Inc.’s charitable status. In that letter, I reviewed a number of developments which supported the conclusion that OpenAI is no longer serving its public, nonprofit purpose and is instead effectively controlled by the for-profit OpenAI affiliate. If OpenAI, Inc. is no longer serving its public, nonprofit purpose, then under California law it should be dissolved, with the value of its assets transferred to another charitable enterprise, such as one or more foundations devoted to artificial intelligence ethics and safety.
I am writing today with supplementary information supporting the theory that OpenAI has abandoned its nonprofit mission.
1. OpenAI’s venture investment in the AI robotics company Figure
On February 29, the AI robotics company Figure announced a major new round of financing – $675 million – that implied a start-up valuation of $2.6 billion. OpenAI Startup Fund is one of the handful of investors in this new round of financing.[1] OpenAI Startup Fund is reported to have invested $5 million.[2] In conjunction with this investment, Figure and OpenAI are entering “into a collaboration agreement to develop next generation AI models for humanoid robots, combining OpenAI’s research with Figure’s deep understanding of robotics hardware and software. The collaboration aims to help accelerate Figure’s commercial timeline by enhancing the capabilities of humanoid robots to process and reason from language.”[3] It is unclear what financial benefits flow to OpenAI from the collaboration agreement, but given the aim to “accelerate Figure’s commercial timeline,” they are likely to be consequential.
There is of course a great deal of uncertainty around any early-stage investment, but the commercial ambitions of Figure and its investors are clear. Figure and its competitors “are racing to cement their place as an early dominant force in an industry some supporters believe could be a real money-maker in the near-future,” reports Popular Science. “In 2022, Goldman Sachs predicted the global humanoid robot market could reach $154 billion by 2035. If that sounds like a lot, it’s a fraction of the $3 trillion financial services company Macquarie estimates the industry could be worth by 2050.”[4]
It’s entirely probable that the executives and employees at Figure and their collaborators at OpenAI believe that marketing and commercializing humanoid robots will serve the public. But it is clear that this is a for-profit enterprise driven by for-profit considerations. The OpenAI investment and partnership with Figure does not have the hallmarks of a nonprofit venture. That may be expected – the OpenAI entities involved in this partnership are almost certainly for-profit themselves – but the governing role of the nonprofit OpenAI and determination of adherence to nonprofit purpose is nowhere to be found.
2. OpenAI, Inc. CEO Sam Altman is the sole owner of OpenAI Startup Fund
In February, Axios reported that OpenAI, Inc. CEO Sam Altman is the sole owner of OpenAI Startup Fund.[5]
This is a confounding and bewildering situation:
- OpenAI told Axios that this situation was intended to be “temporary,” but it has persisted for more than a year.
- OpenAI Startup Fund is investing substantial money – declared to be $175 million – but OpenAI itself is not an investor in OpenAI Startup Fund. Microsoft is among the fund’s investors.[6]
- OpenAI, however, does manage the fund, as Sam Altman states in an announcement video on OpenAI Startup Fund’s website. [7]
- The intermingling of OpenAI and OpenAI Startup Fund is further evidenced by the Figure investment – with the monetary contribution provided by OpenAI Startup Fund and the research and development collaboration seemingly involving OpenAI.
It is very hard to square this set of facts with the idea that the nonprofit OpenAI remains committed to its mission. Instead, they provide compelling evidence that the OpenAI network of enterprises is not truly functioning under the control of the nonprofit OpenAI, Inc. and that the CEO of the nonprofit is oriented to the for-profit mission of the for-profit affiliates, not a nonprofit purpose.
(To be clear, we are not making claims about Altman’s personal monetary interests or potential issues of inurement – we do not have a window into Altman’s compensation, OpenAI ownership interests or relationship with OpenAI partners.)
3. Insights from the Musk lawsuit
On March 1, Elon Musk, CEO of multiple tech companies and one of the original founders of OpenAI, Inc., sued OpenAI, Inc., arguing that the nonprofit violated various commitments made during the founding of the company and to him as the major donor to the nonprofit in its early years.[8] The crux of Musk’s lawsuit is the same as the question we raised in our January 9 letter to you: OpenAI, Inc. appears to have abandoned its nonprofit purpose. Musk’s complaint offers various insights that further support the conclusion that OpenAI, Inc.’s nonprofit purpose has been subordinated to the profit-seeking purposes of the affiliates over which it purportedly exercises control.
First, with the release of ChatGPT-4, OpenAI departed from its prior practice of detailing the training and other internal details of its models. “GPT-4 is an entirely closed model,” according to the Musk complaint. “The internal design of GPT-4 remains a secret and no code has been released. OpenAI has not published a paper describing any aspect of its internal design; it has simply issued press releases boasting about its performance.” This is not just an unfortunate policy change, Musk argues, but one that signals the abandonment of nonprofit purpose: “The internal details of GPT-4 are known only to OpenAI and, on information and belief, to Microsoft. GPT-4 is hence the opposite of ‘open AI.’ And it is closed for propriety commercial reasons.”[9]
Further, Musk contends, “OpenAI’s entire development is now veiled in secrecy and the public only has rumors and isolated fragments of communications to understand what may be released next.” Again, the important point is that this relative secrecy seems to reflect a profit orientation.[10]
Second, Musk offers insights about the change in board membership of OpenAI, Inc. We argued in our January 9 letter that the removal of the old board members during OpenAI’s tumult in November provided strong evidence that for-profit forces had defeated those devoted to the nonprofit’s purpose – a conclusion widely shared among industry watchers.
Musk supplements this claim by contending that the “new Board consist[s] of members with more experience in profit-centric enterprises or politics than in AI ethics and governance,” meaning they may elevate the for-profit interests in the OpenAI network at the expense of OpenAI’s nonprofit mission. The new board members, Musk contends, “are also reportedly ‘big fans of Altman.’ Two of the new board members were Bret Taylor and Larry Summers. Mr. Taylor is no stranger to Silicon Valley, and has been heavily involved in various Bay Area profit-driven ventures. On February 14, 2024, Mr. Taylor and ex-Google executive Clay Bavor launched a startup focused on building AI chatbots for businesses. Dr. Summers is an economist who, on information and belief, had no experience working in artificial intelligence-based ventures prior to November 2023.”[11]
Third, Musk underscores a key power of the OpenAI, Inc. board, central to its nonprofit purpose: It has the authority to determine when OpenAI has developed artificial general intelligence (AGI). OpenAI’s partnership with Microsoft “leave[s] AGI technologies and governance for the Nonprofit and the rest of humanity.”[12] A board with less technical expertise, less commitment to AI ethics and OpenAI, Inc.’s nonprofit purpose, and more oriented toward profit – as the current board is, as compared to the old – will be less able to make determinations about AGI and less inclined to do so in ways that conflict with profit opportunities.
Musk’s complaint explains: “The OpenAI, Inc. Board’s technical expertise in AI, neutrality, and commitment to OpenAI, Inc.’s nonprofit mission are particularly essential to OpenAI, Inc.’s mission as it is the Board that determines whether OpenAI has attained AGI for purposes of its license agreement with Microsoft. This means that the Board is tasked with determining whether OpenAI’s most powerful and advanced technology is actually excluded from the scope of Microsoft’s exclusive license. Given Microsoft’s enormous financial interest in keeping the gate closed to the public, OpenAI, Inc.’s new captured, conflicted, and compliant Board will have every reason to delay ever making a finding that OpenAI has attained AGI.”[13]
One does not have to reach an opinion on how close or far OpenAI may be from AGI to absorb the importance of this point. That is particularly so given that many in and around OpenAI and Microsoft themselves believe that OpenAI to have achieved or be near achieving AGI, as the Musk complaint demonstrates.
There is a final important point to be made here, in light of the Washington Post report that a new OpenAI, Inc. board will soon be appointed.[14] Even if the new board consists of people with demonstrated expertise and commitment to AI ethics and safety, they will operate in the shadow of the November purge and the staff threat to leave OpenAI and seek employment with Microsoft – and with Microsoft apparently maintaining an “observer” seat on the board. It is very hard to imagine any future board re-imposing the nonprofit purpose at this point.
* * *
Thank you for your considering this additional information, which we think further solidifies the view that OpenAI is no longer operating in accordance with its nonprofit mandate. With OpenAI reportedly having secured a deal that values the for-profit at more than $80 billion and will yield substantial rewards for employees,[15] we urge you to investigate OpenAI, Inc.’s nonprofit status expeditiously.
Sincerely,
Robert Weissman,
President, Public Citizen
Endnotes
[1] “Figure Raises $675M at $2.6B Valuation and Signs Collaboration Agreement with OpenAI,” February 29, 2024, https://www.prnewswire.com/news-releases/figure-raises-675m-at-2-6b-valuation-and-signs-collaboration-agreement-with-openai-302074897.html.
[2] “Bezos, Nvidia join OpenAI in funding humanoid robot startup, Bloomberg reports,” Reuters, February 23, 2024, https://www.reuters.com/technology/bezos-nvidia-join-openai-funding-humanoid-robot-startup-bloomberg-reports-2024-02-23/
[3] “Figure Raises $675M at $2.6B Valuation and Signs Collaboration Agreement with OpenAI,” February 29, 2024, https://www.prnewswire.com/news-releases/figure-raises-675m-at-2-6b-valuation-and-signs-collaboration-agreement-with-openai-302074897.html.
[4] Mack Degeurin, “OpenAI Wants to Make a Walking, Talking Humanoid Robot Smarter,” Popular Science, February 29, 2024, https://www.popsci.com/technology/openai-wants-to-make-a-walking-talking-humanoid-robot-smarter/
[5] Dan Primack, “Sam Altman Owns OpenAI’s Venture Capital Fund,” Axios, February 15, 2024, https://www.axios.com/2024/02/15/sam-altman-openai-startup-fund
[6] https://www.openai.fund/about
[7] https://www.openai.fund/about
[8] Elon Musk v. Samuel Altman, et. al. Complaint for breach of contract and other counts, Superior Court of California in and for the County of San Francisco, February 29, 2024, https://www.courthousenews.com/wp-content/uploads/2024/02/musk-v-altman-openai-complaint-sf.pdf.
[9] Musk complaint at paragraph 90.
[10] Musk complaint at paragraph 90.
[11] Musk complaint at paragraph 110.
[12] https://openai.com/our-structure.
[13] Complaint at 112
[14] Gerrit De Vynck and Nitasha Tiku, “OpenAI, facing SEC probe, to name new board members within a month,” The Washington Post, February 29, 2024, https://www.washingtonpost.com/technology/2024/02/29/sec-openai-investors-board-sam-altman
[15]Cade Metz and Tripp Mickle, “OpenAI Completes Deal That Values the Company at $80 Billion,” New York Times, February 16, 2024, https://www.nytimes.com/2024/02/16/technology/openai-artificial-intelligence-deal-valuation.html